An Empirical Analysis of Students' Difficulties on Learning Conceptual Data Modeling
I-Lin Huang · Academy of Information and Management Sciences journal · 2012
ABSTRACT Conceptual data modeling is an error prone process, especially for student database modelers. In order to improve the accuracy of data models, the entity-relationship modeling has been accepted by academics and practitioners as an effective technique to help database modelers to capture, understand, and represent business data requirements. However, the cognitive abilities of database modelers are still the most important determinants for the accuracy of data models. Empirical studies have showed that the performance of student database modelers is significantly lower than that of expert database modelers. The results are well-expected because student database modelers are obviously inadequate in modeling knowledge of the data modeling technique. However, empirical studies have also shown that the learning process is slow for student database modelers to reach the expert level of reasoning processes and modeling performance. This result poses an unanswered question: why student database modelers cannot have significantly better performance of data modeling once they have learned the modeling knowledge? After reviewing the literature on conceptual modeling, this research proposes that in addition to modeling knowledge, two cognitive variables, domain knowledge and cognitive fit; also significantly influence the modeling performance of student database modelers. An experiment is then conducted to test the influence of domain knowledge and cognitive fit on the modeling performance of student database modelers. On the basis of the research results, this research suggests that an effective instruction system for training student data modelers needs to consider the influence of domain knowledge and cognitive fit. INTRODUCTION Statistics show that database administration is one of five fastest growth areas in the job market for the first decade of the new millennium (Newsweek, 1999, page 44, New York Times, 2001). The growth of the demand for competent database modelers reflects the importance of the quality of database systems in supporting wide-spreading e-businesses and enterprise resource planning systems in current business environments (Antony & Batra, 2002). The importance of the quality of database systems in current business environments has also been demonstrated by the vast expense of businesses in erroneous data due to poorly designed databases. According to an estimation, the erroneous customer data alone cost businesses on a global level ranging from under $100 billion to over $600 billion, not to mention the erroneous data in other aspects of businesses (Hillard, McClowry, & Na, 2007). An accurate data model is essential to building a well-functioning database. In developing a database, the data model provides a blueprint and foundation for the structure of the database by showing an abstract representation of the data about entities, their associations and attributes within the intended business (Topi & Ramesh, 2002). If the data model is flawed, the quality of the resulting database will be compromised. However, conceptual data modeling is an error prone process, especially for student database modelers (Antony & Barra, 2002; Batra, 2005; Barra and Sein; 1994; Sutcliffe and Maiden, 1992). In order to improve the accuracy of data models, the entity-relationship modeling has long been accepted by academics and practitioners as an effective technique to help database modelers to capture, understand, and represent business data requirements (Antony & Batra, 2002; Batra and Davis, 1992; Batra and Sein; 1994). However, the cognitive abilities of database modelers are still the most important determinants for the accuracy of data models. It has been reported that student database modelers literally follow the stated requirements to specify entity-relationship models (Batra & Antony, 1994). As a result, the data relationships that are not expressed obviously in requirement statements become the main source of modeling errors committed by students (Batra and Davis, 1992; Batra and Sein; 1994; Sutcliffe and Maiden, 1992). …